Autonomous Stock Screening with AI Scoring
Learn how MiroMint’s AI algorithms process technical indicators, fundamental balance sheet metrics, and news sentiment to generate real-time stock scores.
🎯 Key Learning Objectives
- ✓Traditional screeners filter stocks based on static thresholds, whereas AI screeners analyze non-linear relationships across hundreds of variables.
- ✓MiroMint’s scoring engine evaluates three core vectors: Fundamentals, Technicals, and Alternative Sentiment.
- ✓Natural Language Processing (NLP) parses financial news and SEC filings in real-time to generate a dynamic sentiment vector.
- ✓AI scores are updated continuously to adapt to changing macro environments and stock price fluctuations.
The Evolution of Stock Screening
For decades, investors screened stocks using simple, rule-based filters. For example: "Show me stocks with a P/E ratio under 15, an RSI under 30, and a debt-to-equity ratio under 0.5."
While these filters are useful, they suffer from two major limitations: 1. Static Thresholds: A stock with a P/E of 15.1 is excluded, even if its other metrics are spectacular. 2. Linear Assumptions: Traditional filters cannot evaluate complex, non-linear relationships between variables (e.g., how a sudden rise in bond yields affects growth stock valuations in the semiconductor sector).
AI-powered stock screening solves this by evaluating hundreds of financial and technical parameters simultaneously, using neural networks and decision trees to identify complex patterns.
---
MiroMint’s Multi-Vector Scoring Architecture
MiroMint’s proprietary AI engine scores every stock on a scale of 0 to 100. This score represents the statistical probability that the stock will outperform its benchmark index over the next 30 to 90 days. The engine processes data across three distinct vectors:
\\\
┌──────────────────────────────┐
│ MiroMint AI scoring Pipeline │
└──────────────┬───────────────┘
│
┌───────────────────────┼───────────────────────┐
▼ ▼ ▼
┌──────────────┐ ┌──────────────┐ ┌───────────────┐
│ Fundamentals │ │ Technicals │ │ Alt Sentiment │
│ PE, FCF, │ │ RSI, MACD, │ │ News, Filings │
│ Balance Sheet│ │ Volatility │ │ (NLP Vectors) │
└───────┬──────┘ └──────┬───────┘ └───────┬───────┘
│ │ │
└──────────────────────┼────────────────────────┘
│
▼
┌───────────────────────┐
│ Multi-Vector Engine │
└───────────┬───────────┘
│
▼
┌─────────────────────┐
│ Final AI Score 0-100│
└─────────────────────┘
\\\
1. The Fundamental Vector This vector analyzes the structural health of the company. It ingests quarterly and annual SEC filings (10-K, 10-Q) and evaluates: - **Solvency:** Debt-to-Equity, Interest Coverage, Quick Ratio. - **Profitability:** Gross Margin, Return on Equity (ROE), Free Cash Flow margins. - **Valuation:** Relative ranking of P/E, PEG, Price-to-Sales, and Enterprise Value multiples against sector peers.
2. The Technical Vector This vector captures market momentum and price action. It continuously monitors: - **Trend Strength:** Exponential moving average alignment (20, 50, 200 EMA). - **Momentum Oscillators:** RSI, MACD, Stochastic. - **Volatility Parameters:** Average True Range (ATR), Bollinger Band width, and systematic Beta.
3. The Alternative Sentiment Vector (NLP) Perhaps the most powerful component of modern quantitative terminal architectures is the sentiment engine. MiroMint's AI processes: - **Financial News:** Scanning articles from Bloomberg, Reuters, and Yahoo Finance. - **SEC Management Discussions:** Parsing the textual commentary in quarterly filings to spot subtle shifts in executive tone. - **Earnings Call Transcripts:** Analyzing Q&A sessions using Natural Language Processing (NLP) models to score negative or positive semantic structures.
---
Explainable AI (XAI) in Finance
A common criticism of machine learning in finance is the "black box" problem: a model outputs a buy signal, but provides no explanation of why it did so. This is unacceptable for institutional investors.
To resolve this, MiroMint integrates SHAP (SHapley Additive exPlanations) values into its prediction engine. SHAP values mathematically measure the contribution of each financial metric to the final AI score. If a stock receives an 88 score, our interface shows exactly how much of that score was driven by its strong Q1 earnings beat versus its bullish moving average crossover, ensuring transparency and trust.